7-Day Surge Over 250%! AI Large Models Completely Explode! Global Large Models "Chasing Each Other"

The “C position” on the first day of the Year of the Horse in Hong Kong stocks has been overtaken by large models.

On February 20, the first trading day of the Year of the Horse in Hong Kong stocks, Zhipu, known as the “world’s number one large model stock,” maintained strong momentum and performed even better than before the holiday, closing up over 42% with a market capitalization surpassing HKD 320 billion. Looking at the longer term, since February 9, Zhipu has gained more than 250% over seven trading days. Besides Zhipu, another AI large model company, MiniMax, also continued its pre-holiday strength, closing up over 14%, with a cumulative increase of over 110% in the past seven trading days, and a market cap exceeding HKD 300 billion.

While the market value of listed large model companies soars, numerous unlisted large model enterprises at home and abroad are also announcing new rounds of financing, with OpenAI’s latest funding round expected to exceed $100 billion. Including this round, OpenAI’s overall valuation may surpass $850 billion, a 70% increase since the end of last year. Domestically, the valuation of Kimi, a large model company, has doubled in its latest funding round.

“The Twin Giants of Large Models” Show Impressive Growth

Since February, the “Twin Giants” Zhipu and MiniMax have shown remarkable growth. As of the close on February 20, Zhipu’s cumulative increase in February exceeded 220%, and MiniMax’s stock price has risen over 105% since February.

The driving force behind these surges is likely their latest advancements in large models.

Recently, Zhipu launched its new flagship model, GLM-5, which is positioned as “a foundational model for complex system engineering and long-term agent tasks,” aimed at handling complex programming and intelligent agent tasks. GLM-5 has achieved alignment in programming capabilities comparable to Claude Opus 4.5, outperforming Gemini3 Pro. In the authoritative Artificial Analysis ranking, GLM-5 ranks fourth globally and first among open-source models, setting a new height for domestic models on the global large model stage.

Additionally, on February 12, Zhipu announced a structural adjustment to its GLM Coding Plan pricing, removing the initial purchase discount and increasing overall package prices by at least 30%. This marks the first significant price hike for domestic large models in recent years. The GLM series models are highly sought after by developers worldwide, with user numbers surging. Subscription packages based on these models are now in high demand—an industry first. GLM-5 has also been deeply integrated with domestic computing platforms such as Huawei Ascend, Moore Threads, Cambrian, Kunlun Chip, Muxi, Suiyuan, and Hygon, achieving high throughput and low latency on domestic chip clusters.

Furthermore, Zhipu’s pursuit of a listing on the A-share market has made new progress. On February 9, Guotai Huitong signed a guidance agreement with Zhipu, joining the previously signed guidance agreement with China International Capital Corporation to assist Zhipu’s listing on the STAR Market.

MiniMax is also formidable. Alongside Zhipu’s launch of GLM-5, it officially released its latest flagship programming model, MiniMax M2.5. As the world’s first production-level model designed natively for agent scenarios, its coding and agentic performance rivals top international models, directly comparable to Claude Opus 4.6. It supports full-stack programming development across PC, app, and cross-platform applications, excelling particularly in advanced Excel processing, in-depth research, and PowerPoint—leading the industry in core productivity scenarios (SOTA). The M2.5 model has only 10 billion parameters, with clear advantages in memory usage and inference efficiency, supporting 100 TPS high throughput, with inference speeds far surpassing top international models.

Additionally, the robot that “dominated” the Spring Festival Gala may further boost the momentum of large model concept stocks. In the 2026 Spring Festival Gala, the number and depth of robot performances set new records. The integration of intelligent agents and embodied intelligence is a necessary path toward AGI and a key to unlocking the market potential of embodied intelligence. Wang Xingxing, founder and CEO of Yushu Technology, stated, “Currently, the biggest driving force behind the robot industry or embodied intelligence industry is largely AI-driven, but the growth curve may become steeper and steeper. If in the next few years, we see truly large-scale applications of embodied intelligent AI models and breakthroughs in robot technology, the heat could be at least 100 times, even 1,000 times higher than now. I believe this excitement will far surpass mobile internet. So, I think the current stage is definitely not the hottest; we are probably still in the ramp-up or platform stage.”

Global Competition in Large Models Intensifies

Since 2026, the global large model field has seen fierce competition, with major tech giants like Alibaba, ByteDance, Tencent competing to “spend money” to attract C-end users; meanwhile, new rounds of financing have been announced by domestic and international large model companies.

According to reports, Kimi, a domestic large model company, completed a $500 million funding round just over a month after its previous round, with a new round of over $700 million nearing completion, led jointly by Alibaba, Tencent, Wuyuan, and Jiuan. The pre-money valuation for this round is close to $4.8 billion, while at the end of December 2025, Kimi announced a Series C funding of $500 million with a post-money valuation of $4.3 billion. A new funding round valuing the company at $10-12 billion has also begun, with multiple institutions expressing interest, rapidly doubling its valuation past $10 billion.

Overseas, AI startup Anthropic recently completed a $30 billion financing, with a post-money valuation of $380 billion. Meanwhile, OpenAI is close to completing the first phase of a new funding round expected to raise over $100 billion. Sources say that including this round, the company’s overall valuation could exceed $850 billion, higher than previous estimates of $830 billion. By the end of 2025, OpenAI’s valuation is projected to reach about $500 billion.

Recently, “AI godmother” Fei-Fei Li’s startup World Labs announced it had completed a new $1 billion funding round. World Labs focuses on building world models to enhance spatial intelligence, aiming to revolutionize narrative creation, creative design, robotics, and scientific discovery. At the end of last year, the company launched its first product, Marble, which can generate 3D worlds from images or text.

According to Sullivan Data, by 2030, AI is expected to empower at least 20% of daily business decisions globally, support 80% of mainstream consumer smart devices, and create an economic impact of over $20 trillion. China’s AI market size grew from 93.7 billion yuan in 2022 to 160.7 billion yuan in 2024, with a compound annual growth rate of 31.0%. It is projected to further expand to 993 billion yuan by 2030, with a CAGR of 35.5% from 2024 to 2030.

Guosen Securities research indicates that as more Agent products are rapidly deployed domestically and internationally, AI is gradually changing the way users interact with software and the internet. Large models are beginning to create value, and it is recommended to focus on the most certain computing power and large model companies.

(Article source: Securities Times)